El Niño chances rising for hurricane season

Sea surface temperatures in the Equatorial Eastern Pacific have been rising steadily for several months, and there is now a very real possibility that an El Niño event could occur during the peak of the Atlantic hurricane season, August - October. This is important, since the number and intensity of Atlantic tropical storms and hurricanes is usually reduced during an El Niño year, thanks to the increased wind shear such events bring to the tropical Atlantic. Last month, Columbia University's International Research Institute (IRI) was giving a 30% chance of an El Niño event for the coming hurricane season; this month, they have bumped their odds up to 45%. The Australian Bureau of Meteorology notes that "recent trends are consistent with the very early stages of a developing El Niño". NOAA's Climate Prediction Center forecasts the current neutral conditions in the Equatorial Eastern Pacific will continue into the summer, but shows that their CFS El Niño model is predicting a moderate El Niño event for the coming hurricane season.


Figure 1. Sea Surface Temperature (SST) departure from average for the equatorial Eastern Pacific (the area 5°N - 5°S, 120°W - 170°W, also called the "Niña 3.4 region"). The +0.5°C mark is the threshold for El Niño conditions, and we are very close to that mark now. Image credit: NOAA's Climate Prediction Center.

El Niño forecast models
The latest suite of runs by the various computer models used to forecast El Niño offer two main forecasts for the coming hurricane season: neutral conditions will persist, or an El Niño will develop (Figure 2). There are two types of models used to make these forecasts: statistical models and dynamical models. The statistical models have been around the longest, and they rely on statistics of how past El Niño episodes have developed in order to make a forecast. Dynamical models don't care about what has happened in the past. They make a forecast by taking the current state of the atmosphere, putting the data on a grid covering the entire globe, then solving the equations that govern the physics of the atmosphere and ocean on each point of this grid every few minutes, marching forward in time for many months. These dynamical models, in many cases, are simply modified versions of the same models we use to forecast the short-term weather. For example, the NOAA's Coupled Forecast System (CFS) model is based on the GFS model that we use to track hurricanes and make short-term weather forecasts. The main difference is that the CFS model runs for many months instead of just a few days.


Figure 2. El Niño model forecasts made in mid-May. Note that for the peak part of hurricane season, August-September-October (ASO), most of the dynamical models are forecasting an El Niño event (SST anomaly greater than 0.5°C in the Eastern Equatorial Pacific). Image credit: Columbia University's IRI.

Which model to believe?
As is the case with all seasonal forecasts, El Niño forecasts are not very good, and don't do much better than flipping a coin. However, thanks to intensive research efforts and the doubling of computer power that has been occurring every 1.5 years, the El Niño forecasts by the dynamical models have improved considerably over the past few years. These models now do about as well as the traditional statistical models, and should continue to improve as computer power continues to increase and our understanding of El Niño increases. Over the past two months, the dynamical models have increasingly been forecasting the development of an El Niño this Fall. To illustrate, in March only three of the thirteen dynamical models were predicting an El Niño event for hurricane season. By mid-May, this had increased to nine out of thirteen models. However, none of the eight statistical El Niño models are forecasting an El Niño event for the Fall, and their forecasts should be respected, as well. The IRI web site has a nice tool one can use to study the performance of the individual models. To my eye, the Japanese Meteorological Agency (JMA) dynamical model has made the best El Niño forecasts over the past two years (though I haven't done a rigorous error analysis to verify this). The JMA model is predicting a weak El Niño event for the coming hurricane season, and I am going to go along with that forecast.

What will an El Niño event do to hurricane numbers?
Since the active hurricane period we are in began in 1995, there have been four El Niño events (Figure 3). During these years, the number of named storms, hurricanes, and intense hurricanes 11 named storms, 5 hurricanes, and 3 intense hurricanes. This is close to the average levels we've seen over the past 60 years--10-11 named storms, 6 hurricanes, and 2 intense hurricanes. If, on the other hand, we look at the five years that had neutral conditions, the numbers are considerably higher--18 named storms, 10 hurricanes, and 5 intense hurricanes. So, let's hope for an El Niño this year. Note, though, that one of our worst hurricane years--2004, which featured hurricanes Ivan, Charlie, Frances and Jeanne, which all affected Florida with hurricane conditions--was an El Niño year. It seems that in years like 2004, there is a lag between the time a El Niño event develops and the response of the atmosphere over the Atlantic. There is no way of forecasting at this point whether this could be the case this year. One argument against a repeat of 2004 is the presence of much lower heat content and SSTs in the tropical Atlantic this year compared to 2004.


Figure 3. Looking at the numbers of Atlantic names storms, hurricanes, and intense hurricanes since 1995.

Tropical Depression One
The season's first tropical depression formed yesterday off the coast of North Carolina, but has missed its opportunity to become the Tropical Storm Ana. Tropical Depression One is headed east-northeastward out to sea, and is now entering a region with cooler water temperatures and increased wind shear of 15 - 20 knots. The heavy thunderstorm activity associated with TD One has shrunk this morning, and high wind shear has pushed these thunderstorms to the east side of the center, exposing the surface circulation to view. Tropical Depression One will not hold together much longer, and should be history by Saturday night.

Is the formation of TD One a harbinger of an active hurricane season?
Probably not. Early season storms occurring near the U.S. coast have not been shown to be correlated with an active main portion of hurricane season during August - October. However, the situation is different if we start getting June and July storms in the deep tropics between Africa and the Lesser Antilles Islands. This was the case last year, when the formation of Hurricane Bertha in the deep tropics in July presaged an active 2008 hurricane season. According to the Hurricane FAQ, "as shown in (Goldenberg 2000), if one looks only at the June-July Atlantic tropical storms and hurricanes occurring south of 22°N and east of 77°W (the eastern portion of the Main Development Region [MDR] for Atlantic hurricanes), there is a strong association with activity for the remainder of the year. According to the data from 1944-1999, total overall Atlantic activity for years that had a tropical storm or hurricane form in this region during June and July have been at least average and often times above average. So it could be said that a June/July storm in this region is pretty much a "sufficient" condition for a year to produce at least average activity."

I'll have a detailed outlook of the coming hurricane season on Monday, the first day of the Atlantic hurricane season.

Jeff Masters